Topic hub pagination
Continue browsing this topic cluster with SEO-safe static pagination.
Developers using LiteLLM as proxy encountering timeouts but unable to determine whether failure is network, provider-side, or configuration-related; need timeout debugging guidance. Includes evidence for LiteLLM troubleshooting demand.
TimeoutError: Request timed out — LiteLLM proxy swallows provider-level timeout without exposing root cause Developer configuring LiteLLM as API proxy needs to resolve rate limiting, connection timeouts, and provider routing errors affecting their unified LLM endpoint Includes evidence for LiteLLM troubleshooting demand.
LiteLLM proxy returns timeout, rate limit exceeded, and model routing failures when forwarding requests to upstream LLM providers Fix LiteLLM proxy fallback configuration not working for team-scoped BYOK models due to public vs internal model name mismatch Includes evidence for LiteLLM troubleshooting demand.
Fallback lookup fails for team-scoped models — model_group uses public name but fallback config requires internal names Fix LiteLLM proxy timeout or backend overloaded errors when acting as a single entry point for multiple LLM provider backends Includes evidence for LiteLLM troubleshooting demand.
LiteLLM proxy timeout error / backend overloaded_error when routing requests to upstream LLM providers Fix LiteLLM proxy rate limiting, request timeouts, and unknown model errors blocking AI API routing for production workloads Includes evidence for LiteLLM troubleshooting demand.
LiteLLM proxy: RateLimitError / timeout error / model_not_found — requests to 100+ LLM APIs fail due to misconfigured RPM/TPM limits or missing model_list entries Fix LiteLLM proxy passing invalid JSON Schema fields (title, default) to Fireworks AI when bridging MCP tools, causing API rejections for Coralogix/DevRev integration users Includes evidence for LiteLLM troubleshooting demand.
Fireworks AI rejects MCP tool schemas unsupported JSON Schema title default litellm Using LiteLLM as proxy with MCP tool calling, Fireworks AI provider rejects tool schemas containing non-standard JSON Schema fields (title, default), breaking Coralogix and DevRev MCP integrations Includes evidence for LiteLLM troubleshooting demand.
Fireworks AI rejects MCP tool schemas with unsupported JSON Schema fields (title, default) — triggered by Coralogix & DevRev MCP integrations Using LiteLLM virtual keys with token-per-minute/rate-per-minute limits causes internal _litellm_* parameters to be leaked into upstream provider API calls, corrupting requests and breaking automatic fallback routing Includes evidence for LiteLLM troubleshooting demand.
Virtual keys with tpm/rpm limits leak _litellm_* params into provider API calls, breaking fallback chains Fix LiteLLM chatgpt provider where database-registered models bypass ChatGPTResponsesAPIConfig causing streaming requests to fail with explicit error 'Stream must be set to true', breaking integration between Claude Code and OpenRouter-hosted ChatGPT-compatible endpoints Includes evidence for LiteLLM troubleshooting demand.
chatgpt provider: DB-registered models skip ChatGPTResponsesAPIConfig — streaming requests fail with 'Stream must be set to true' Users proxying long-running requests through LiteLLM face premature 60s kill; need to properly configure timeout propagation to underlying provider. Includes evidence for LiteLLM troubleshooting demand.
LiteLLM kills request after 60s default timeout despite router_settings.timeout being set higher Developers using LiteLLM proxy need to fix retry mechanism that doesn't respect rate limit headers or configured delays, causing cascading 429 errors. Includes evidence for LiteLLM troubleshooting demand.
litellm.RateLimitError: Rate limit exceeded without proper retry delay Recover from or prevent damage caused by LiteLLM npm package supply chain compromise affecting production LLM routing pipelines Includes evidence for LiteLLM troubleshooting demand.
npm package tampering (supply chain attack) breaking LiteLLM gateway used by CrewAI, DSPy, Browser-Use, Opik and nearly every major AI agent framework Fix overly aggressive retry policy in LLM middleware that causes repeated 429 errors when exceeding provider rate limits, because max backoff duration is shorter than provider's rate limit window Includes evidence for LiteLLM troubleshooting demand.
LLM retry policy too aggressive for provider rate-limit (429) windows — Gemini quota windows exceed max backoff Fix LiteLLM proxy entering infinite health check loop after setting background_health_checks: true, blocking model provisioning and API availability Includes evidence for LiteLLM troubleshooting demand.
[Bug]: LiteLLM proxy enters infinite loop when background_health_checks is set to true Understand why LiteLLM proxy applies unexpected rate limits when keys are in teams, and how to properly configure per-model rate limiting overrides Includes evidence for LiteLLM troubleshooting demand.
LiteLLM rate limit exceeded when key belongs to team; resolution order: Key metadata > Key model_max_budget > Team metadata Fix LiteLLM token_counter crashing with ValueError when optional image_url field is null in multimodal messages, causing billing calculation to fail Includes evidence for LiteLLM troubleshooting demand.
ValueError: Error getting number of tokens from content list: Invalid image_url type: NoneType. Expected str or dict with 'url' field., default_token_count=None Fix LiteLLM proxy errors when calling Anthropic adaptive-thinking models, where LiteLLM incorrectly forces tool_choice parameter and causes API requests to be rejected by Anthropic. Includes evidence for LiteLLM troubleshooting demand.
LiteLLM forces tool_choice parameter on adaptive-thinking Anthropic models (Claude Opus 4.7 etc.) causing request rejection — fix applied in pull #28114 Diagnose and fix OpenRouter model routing errors when models return unclear failure messages, affecting production deployments using OpenRouter as a paid API proxy layer. Includes evidence for OpenRouter troubleshooting demand.
OpenRouter model routing returns ambiguous error when requested model is unavailable or disabled, without distinguishing between temporary unavailability and permanent deprecation Implement fair queuing in LiteLLM proxy to prevent single client from monopolizing model capacity and causing starvation for other tenants Includes evidence for LiteLLM troubleshooting demand.
No fair queuing — traffic burst from one client starves other clients' requests Developer using LiteLLM proxy finds that restricted users can see full model list via GET /v1/models endpoint despite access group restrictions, creating a security/authorization inconsistency Includes evidence for LiteLLM troubleshooting demand.
GET /v1/models ignores user.models restriction — shows all proxy models regardless of user access groups